Database developers, eh? Usually the network, itself, isn't the issue, unless we're talking long distant WAN links, and in those cases, the physical speed of signal transmission can be an issue.
Unfortunately, many developers, database and other kinds, don't fully appreciate what happens "under the covers", and also often don't fully appreciate the possible impact of scale. Often developers work to meet functional requirements to the exclusion of all other considerations.
If a developer has something that works great in their "test/development" environment, but it doesn't work so great outside of that environment, there must be something wrong in the latter's environment.
A teaching example I've used (aimed at relational databases) is my explaining how on some of the early small computers I trained on didn't have multiplication hardware. If you wanted to multiply two numbers, you might do it by adding. So, for example, for 2 times 3, you might add 2 three times or 3 twice.
If, though, you want to multiply 2 by 3,000,000, you could add 2 three million times, or 3,000,000 twice. The latter was much, much quicker than the former.
Relational database joins, are much like the foregoing multiplication examples, when you don't have hardware that can multiply. It doesn't much matter how you join a table/relation of 2 rows/tuples with 3 rows/tuples. If can matter, much, how you join 2 rows/tuples with 3,000,000 rows/tuples. (Query optimizer any one?)
Where additive multiplication, or table/relation joins, become "interesting" is when you need to do 2,000,00 by 3,000,000.
